How to Fix pip install . Failing Due to Missing numpy in Build Isolation

Learn why `pip install .` fails to find numpy during build isolation and how to fix it by managing build dependencies correctly.
---
This video is based on the question https://stackoverflow.com/q/79388211/ asked by the user 'user3144772' ( https://stackoverflow.com/u/3144772/ ) and on the answer https://stackoverflow.com/a/79388523/ provided by the user 'phd' ( https://stackoverflow.com/u/7976758/ ) at 'Stack Overflow' website. Thanks to these great users and Stackexchange community for their contributions.

Visit these links for original content and any more details, such as alternate solutions, latest updates/developments on topic, comments, revision history etc. For example, the original title of the Question was: pip install . fails, looking into /var/private/folders

Also, Content (except music) licensed under CC BY-SA https://meta.stackexchange.com/help/l...
The original Question post is licensed under the 'CC BY-SA 4.0' ( https://creativecommons.org/licenses/... ) license, and the original Answer post is licensed under the 'CC BY-SA 4.0' ( https://creativecommons.org/licenses/... ) license.

If anything seems off to you, please feel free to drop me a comment under this video.
---
The Problem: pip install . Fails but python3 setup.py install Works

When installing a local Python module using pip install ., you might encounter an error where pip cannot find the numpy module, even though it is installed and importable in your environment. Meanwhile, running python3 setup.py install works correctly.

This usually happens for modules that use the Python and NumPy C APIs, with build dependencies declared or imported in setup.py.



Why Does This Happen?

Modern pip uses build isolation, which means it:

Creates a temporary isolated environment to build your package.

Installs build dependencies only inside this temp environment.

Removes the temp environment after the build completes or fails.

Your error arises because:

numpy is needed during the build phase (for include paths via numpy.get_include()).

Pip’s isolated environment does not have numpy installed initially.

Your build fails while trying to import numpy in the isolated temp environment, not your regular Python.

Since the temp directory is cleaned up after the build, you cannot access it.



Two Ways to Fix the Issue

1. Install numpy Globally and Disable Build Isolation

You can install numpy into your active environment first, then tell pip not to use build isolation:

[[See Video to Reveal this Text or Code Snippet]]

This way, the build sees the global numpy installation and can successfully compile your extension.



2. Declare Build-Time Dependencies Properly with pyproject.toml

The recommended, modern approach is to declare build dependencies explicitly using a pyproject.toml file:

[[See Video to Reveal this Text or Code Snippet]]

This informs pip to install numpy automatically into the isolated build environment before running your build.

Key points:

This is the PEP 517/518 standard way to manage build dependencies.

Your setup.py can safely import numpy because it will be installed during build.



Additional Tips

Avoid using setup.py install directly as it is deprecated.

Use pip install . with properly declared build dependencies for consistent results.

Keep your setup.py minimal and clean for smooth integration with modern tools.



Summary

If you encounter ModuleNotFoundError: No module named 'numpy' during pip install ., it’s because pip’s build isolation environment doesn’t have numpy installed.

Fix it by either:

Installing numpy globally and using pip install --no-build-isolation .

Declaring numpy as a build dependency in pyproject.toml under [build-system].

This ensures numpy is available during build time and your installation will succeed seamlessly.